Re: [PATCH 0/5] target: Add se_session->sess_kref + convert iscsi/qla2xxx fabrics

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

On Tue, 28 February 2012 15:15:25 -0800, Nicholas A. Bellinger wrote:
> On Tue, 2012-02-28 at 10:58 -0500, Jörn Engel wrote:
> > On Tue, 28 February 2012 00:56:15 -0800, Nicholas A. Bellinger wrote:
> > > 
> > > The sessions associated with the se_node_acl->acl_group are required to
> > > be shutdown because they may be referencing struct se_node_acl memory,
> > > and the configfs rmdir operation for se_node_acl->acl_group or any
> > > group) is not allowed to fail.
> > 
> > Sounds like we can either add reverse mapping from the configfs
> > structure to all referencing sessions and rip out the references or go
> > through an indirection and simple mark the configfs parts as gone in
> > the indirection.
> We expect to have access to se_node_acl while a se_session is active.
> Ripping out references of se_node_acl from a live se_session is not the
> right thing to do here..
> > I see no reason to tear down and recreate sessions
> > just to fulfill the configfs requirement.
> Sure there is.  With most fabrics (including iscsi-target) when a
> NodeACL is removed it's effectively saying "I don't want this initiator
> to have access anymore to this endpoint".  This involves stopping the
> active session and preventing a new session from being created..

That is a different issue that has nothing to do with configfs, so I
stand by my words.

> tcm_qla2xxx is special because we still allow demo-mode in order to
> prevent issues with FC clients exploding when the fabric login request
> is rejected.

Sounds as if we want a fabric callback here, plus a default
implementation that disables the session (mark it such that no new
commands are accepted, drop the refcount so that it gets freed once
the last command is done).  Then tcm_qla2xxx can go slightly more
funky and fulfill the same requirement of no longer accessing the ACL
after returning from this callback.

To unsubscribe from this list: send the line "unsubscribe target-devel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at

[Linux SCSI]     [Kernel Newbies]     [Linux SCSI Target Infrastructure]     [Share Photos]     [IDE]     [Security]     [Git]     [Netfilter]     [Bugtraq]     [Photos]     [Yosemite]     [Yosemite News]     [MIPS Linux]     [ARM Linux]     [Linux Security]     [Linux RAID]     [Linux ATA RAID]     [Linux IIO]     [Samba]     [Video 4 Linux]     [Device Mapper]     [Linux Resources]

Add to Google Powered by Linux